By skyforbes 
I've recently gotten into coding, MCP tools, and local AI — so between a local AI PC and Sonnet 4.5 for a bit of heavy lifting with code — I've not needed ChatGPT anymore, as my local AI already has a lot of the functionality I need. So I figured — let's download my data to archive it (and maybe even load it into a local AI memory database 🤷) and move on.
OpenAI have just been… strange about this? Firstly — I make the request, it's taking a while so I email them to check in. Standard response "yeah just working on it will escalate the request and get back to you" kind of thing.
I decide to be cheeky and throw in a little "while I have you, can I also have all of the CoT tokens that are obscured from the user? Australian privacy policy seems to imply I have a right to see them." (actually i wrote it like an armchair lawyer squinting at 471 pages of a 1988 legal document but that was the jist) and obviously I receive ZERO response to my check-in emails after that – i get the data EXACTLY 30 days later. (i think it includes the data anyway just not plainly – i'll have to set a code-bot loose in a virtual environment with the export to find out)
But in looking at the chat.html file… Brother this was exported by some poor OpenAI intern with a third party Chrome addon 2 days ago…
Looking at my data export from August, the chat.html was structured differently — i've yet to look through my conversations.json file (it's 400mb of text in a single line lol) but im curious if that is majorly different too 🤔
So uh…. what's going on, Sam? Why'd you get one of the coffee boys to use a third party exporter for your homework 2 days before the deadline?